Identification and Prediction of Multiple Intelligence Patterns among University Students using Machine Learning for Personalized

Humans possess various cognitive strengths, yet traditional educational systems mainly focus on Linguistic and Mathematical intelligences, overlooking other critical learning patterns. This research aims to identify and predict learning behaviors in university students by analyzing features such as academic performance, study habits, attendance patterns, and engagement levels. A multi-algorithm framework was developed, combining supervised and unsupervised learning techniques, to analyze learner profiles and associated parameters from the Learning Meta-Learning dataset. The study utilized K-Means clustering, Stacking, Voting, Decision Trees, and MLP (Multi-Layer Perceptron) classifiers, with the Stacking Classifier achieving 99.2% accuracy. The model categorized learning patterns into classes like "often," "moderate," "few," and "frequent." The target variable was predicted by incorporating key features such as attendance rates, participation in activities, and prior academic performance. By refining feature selection and improving prediction accuracy, these results contribute to personalized learning strategies. The findings provide a comprehensive framework for understanding student behavior and offer actionable insights to enhance educational practices.

Keywords: Learning patterns, academic performance, attendance patterns, K-Means, Stacking Classifier, MLP, educational personalization, student behavior, feature selection,